Regulating Driving Automation Safety

Abstract

Over forty thousand people die in motor vehicle crashes in the United States each year, and over two million are injured. The careful deployment of driving automation systems could prevent many of these deaths and injuries, but only if it is accompanied by effective regulation. Conventional vehicle safety standards are inadequate because they can only test how technology performs in a controlled environment. To assess the safety of a driving automation system, regulators must observe how it performs in a range of unpredictable, real world edge cases. The National Highway Traffic Safety Administration (NHTSA) is trying to adapt by experimenting with a novel regulatory strategy. Instead of setting standards, the agency is using its statutory powers in unprecedented ways—ordering automation developers to report crashes daily and directing rapid recalls that require changes to defective software. NHTSA is betting that intense monitoring and the credible threat of recalls will push developers to prioritize safety. This Article argues that NHTSA’s experimental strategy could be transformed into effective safety regulation. Regulators should (1) require that all new vehicles be equipped with telematics that can send safety data and receive software updates over the air; (2) mandate universal crash reporting; and (3) use recalls to force developers of driving automation systems that create unreasonable risks to restrict where their systems can operate until they can develop safer code.